International Research Journal of Education and Technology

The expanding environmental footprint of road transport underscores the need for precise tracking and evaluation of vehicle CO₂ emissions. This research employs data-driven methods and machine learning to assess and forecast CO₂ emission ratings for light-duty vehicles. Drawing from a 2022 Canadian vehicle dataset that includes fuel consumption details, CO₂ emissions, emission ratings, and smog scores, we preprocessed and examined the data to build reliable predictive models.

We implemented two supervised machine learning approaches—Random Forest Classifier and Decision Tree Classifier—leveraging critical features like engine size, cylinder count, fuel type, transmission type, and fuel efficiency measures. The Random Forest model delivered 99% accuracy on the test data, with the Decision Tree model close behind at 98%, underscoring their strong performance in emission rating predictions. This study introduces a practical predictive tool for rapid assessment of vehicle environmental impact, empowering policymakers and buyers to opt for greener transport choices. Ultimately, the findings demonstrate machine learning's value in curbing carbon emissions and advancing sustainable transportation.